private void ProcessNextJob_Group(NextJob next_job, string temp_ocr_result_filename)
{
// Check that this PDF has not failed before
string check_failed_group_token = NextJob.GetCurrentJobToken(next_job.job, next_job.is_group);
if (!failed_pdf_group_tokens.Contains(check_failed_group_token))
{
// Build up the page numbers string
string page_numbers_string;
{
int page_range_start = ((next_job.job.page - 1) / next_job.job.TEXT_PAGES_PER_GROUP) * next_job.job.TEXT_PAGES_PER_GROUP + 1;
int page_range_end = page_range_start + next_job.job.TEXT_PAGES_PER_GROUP - 1;
page_range_end = Math.Min(page_range_end, next_job.job.pdf_renderer.PageCount);
StringBuilder sb = new StringBuilder();
for (int page = page_range_start; page <= page_range_end; ++page)
{
sb.Append(page);
sb.Append(',');
}
page_numbers_string = sb.ToString();
page_numbers_string = page_numbers_string.TrimEnd(',');
}
string ocr_parameters =
""
+ "GROUP"
+ " "
+ '"' + next_job.job.pdf_renderer.PDFFilename + '"'
+ " "
+ page_numbers_string
+ " "
+ '"' + temp_ocr_result_filename + '"'
+ " "
+ '"' + ReversibleEncryption.Instance.EncryptString(next_job.job.pdf_renderer.PDFUserPassword) + '"'
+ " "
+ '"' + next_job.job.language + '"'
;
if (CheckOCRProcessSuccess(ocr_parameters))
{
next_job.job.pdf_renderer.StorePageTextGroup(next_job.job.page, next_job.job.TEXT_PAGES_PER_GROUP, temp_ocr_result_filename);
}
else
{
// If the group fails, then we queue it up for single OCR attempts...
string new_failed_group_token = NextJob.GetCurrentJobToken(next_job.job, next_job.is_group);
failed_pdf_group_tokens.Add(new_failed_group_token);
// ... and queue it up for single OCR attempts.
QueueJobSingle(next_job.job);
}
}
else
{
// Queue previously failed attempts on this PDF file for single OCR attempts.
QueueJobSingle(next_job.job);
}
}
/// <summary>
/// NB: This must be called inside the queue_lock!
/// </summary>
/// <param name="nextJob"></param>
private void RecordThatJobHasStarted_LOCK(NextJob next_job, object queue_lock_REMINDER)
{
string token = NextJob.GetCurrentJobToken(next_job.job, next_job.is_group);
HashSet <string> current_jobs = next_job.is_group ? current_jobs_group : current_jobs_single;
if (current_jobs.Contains(token))
{
Logging.Error("Job is already running: {0}", token);
}
current_jobs.Add(token);
}
public bool JobGroupHasNotFailedBefore(Job job)
{
string check_failed_group_token = NextJob.GetCurrentJobToken(job, true);
//Utilities.LockPerfTimer l1_clk = Utilities.LockPerfChecker.Start();
lock (queue_lock)
{
//l1_clk.LockPerfTimerStop();
return(!failed_pdf_group_tokens.Contains(check_failed_group_token));
}
}
private void RecordThatJobHasCompleted(NextJob next_job)
{
lock (queue_lock)
{
string token = NextJob.GetCurrentJobToken(next_job.job, next_job.is_group, false);
HashSet <string> current_jobs = next_job.is_group ? current_jobs_group : current_jobs_single;
if (!current_jobs.Contains(token))
{
Logging.Error("Job is not running, so can't remove it: " + token);
}
current_jobs.Remove(token);
}
}
private void RecordThatJobHasCompleted(NextJob next_job)
{
string token = NextJob.GetCurrentJobToken(next_job.job, next_job.is_group);
Utilities.LockPerfTimer l1_clk = Utilities.LockPerfChecker.Start();
lock (queue_lock)
{
l1_clk.LockPerfTimerStop();
HashSet <string> current_jobs = next_job.is_group ? current_jobs_group : current_jobs_single;
if (!current_jobs.Contains(token))
{
Logging.Error("Job is not running, so can't remove it: {0}", token);
}
current_jobs.Remove(token);
}
}
private bool IsSimilarJobRunning(Job job, bool is_group, object queue_lock_REMINDER)
{
// Check if a similar group job is running
if (current_jobs_group.Contains(NextJob.GetCurrentJobToken(job, true)))
{
return(true);
}
// If this is a single job, check if a similar single job is running
if (!is_group)
{
if (current_jobs_single.Contains(NextJob.GetCurrentJobToken(job, false)))
{
return(true);
}
}
// No similar job is running...
return(false);
}
private void ProcessNextJob_Group(NextJob next_job, string temp_ocr_result_filename)
{
// Check that this PDF has not failed before
if (JobGroupHasNotFailedBefore(next_job.job))
{
// Build up the page numbers string
string page_numbers_string;
{
int page_range_start = ((next_job.job.page - 1) / Job.TEXT_PAGES_PER_GROUP) * Job.TEXT_PAGES_PER_GROUP + 1;
int page_range_end = page_range_start + Job.TEXT_PAGES_PER_GROUP - 1;
page_range_end = Math.Min(page_range_end, next_job.job.pdf_renderer.PageCount);
StringBuilder sb = new StringBuilder();
for (int page = page_range_start; page <= page_range_end; ++page)
{
sb.Append(page);
sb.Append(',');
}
page_numbers_string = sb.ToString();
page_numbers_string = page_numbers_string.TrimEnd(',');
}
string ocr_parameters =
""
+ "GROUP"
+ " "
+ '"' + next_job.job.pdf_renderer.DocumentPath + '"'
+ " "
+ page_numbers_string
+ " "
+ '"' + temp_ocr_result_filename + '"'
+ " "
+ '"' + ReversibleEncryption.Instance.EncryptString(next_job.job.pdf_renderer.PDFPassword) + '"'
+ " "
+ '"' + next_job.job.language + '"'
;
// https://stackoverflow.com/questions/2870544/c-sharp-4-0-optional-out-ref-arguments
if (CheckOCRProcessSuccess(ocr_parameters, out _))
{
next_job.job.pdf_renderer.StorePageTextGroup(next_job.job.page, Job.TEXT_PAGES_PER_GROUP, temp_ocr_result_filename);
}
else
{
// If the group fails, then we queue it up for single OCR attempts...
string new_failed_group_token = NextJob.GetCurrentJobToken(next_job.job, next_job.is_group);
//Utilities.LockPerfTimer l1_clk = Utilities.LockPerfChecker.Start();
lock (queue_lock)
{
//l1_clk.LockPerfTimerStop();
failed_pdf_group_tokens.Add(new_failed_group_token);
}
// ... and queue it up for single OCR attempts.
QueueJobSingle(next_job.job);
}
}
else
{
// Immediately queue previously failed GROUP attempts on this PDF file as SINGLE OCR attempts instead,
// without even trying the GROUP mode again, for it will certainly fail the second/third/etc.
// time around as well.
QueueJobSingle(next_job.job);
}
}